Genomic structure, polymorphism and expression of ACCN1 and ACCN3 genes in the horse


X. Mata, INRA, UMR1313, Unité de Génétique Animale et Biologie Intégrative, Centre de Recherche de Jouy, 78350 Jouy-en-Josas, France.


A category of cation gate proteins was shown to be present in sensory neurons and act as receptors of protons present in tissues such as muscles. The Amiloride-sensitive Cation Channel, Neuronal (ACCN) gene family is known to play a role in the transmission of pain through specialized pH sensitive neurons. Muscles from horses submitted to strenuous exercises produce lactic acid, which may induce variable pain through ACCN differential properties. The sequences of the equine cDNAs were determined to be 2.6 kb in length with an open reading frame of 1539 bp for ACCN1 and 2.1 kb in length with an open reading frame of 1602 bp for ACCN3. The ACCN1 gene is 990 kb long and contains 10 exons, and the ACCN3 gene is 4.2 kb long and contains 11 exons. The equine ACCN1 and ACCN3 genes have an ubiquitous expression but ACCN1 is more highly expressed in the spinal cord. We identified one alternative ACCN3 splicing variant present in various equine tissues. These mRNA variants may encode two different protein isoforms 533 and 509 amino acids long. Ten single nucleotide polymorphisms (SNPs) were detected for ACCN1; five in the coding and five in the non-coding region, with no amino acid change, while the three SNPs identified in the coding region of the ACCN3 gene introduce amino acid changes. The equine in silico promoter sequence reveals a structure similar to those of other mammalian species, especially for the ACCN1 gene.


Pain is caused by multiple factors such as mechanical, thermal or chemical stimuli that are processed by specialized neurons classified according to their characteristics. Those of the cation gates triggering the nerve impulse by depolarization of the cell membrane are particularly involved. The signal originates in the terminal part of neurons because of tissue damage and is conveyed to the brain, where it is interpreted as an uncomfortable perception. But pain is also a source of information for the living organism to react and protect tissue or body from alteration or destruction. The existence of interindividual differences in pain perception suggests the involvement of genetic factors (Mogil 1999, 2004). Pain research has undergone major evolution at the system, cellular, subcellular and molecular levels. One source of chemical stimulation triggering pain is the high concentration of hydrogen ions, which lowering the tissue pH in the cellular environment, and this can be due in part to lactic acid secretion.

A category of cation gates was shown to be present in sensory neurons and act as receptors of acidity. The Amiloride-sensitive Cation Channel, Neuronal (ACCN, also known as Acid Sensing Ion Channel, ASIC) gene family probably plays an important role in the transmission of pain through specialized pH sensitive neurons (for review see Kellenberger & Schild 2002; Mamet & Voilley 2002). The first gene of the family was cloned in a neurodegenerative mutant of Caenorhabditis elegans and was thus called degenerin (Driscoll & Chalfie 1992). To date, several other members of this degenerin/epithelial Na+ channel superfamily have been identified and cloned in mammals. The ACCN genes are differentially distributed in the central and/or peripheral nervous system, and four of them, ACCN1 (ASIC2) (Lingueglia et al. 1997), ACCN2 (Waldmann et al. 1997b; Chen et al. 1998), ACCN3 (ASIC3) (Waldmann et al. 1997a) and ACCN4 (Grunder et al. 2000), have been cloned. In addition, two isoforms are generated by alternative splicing for ACCN2 (Waldmann et al. 1997b; Ugawa et al. 2001) and three for ACCN3. Two initiation sites in ACCN1 trigger the alternative transcription of exon 1a or exon 1b in humans (Lingueglia et al. 1997) and in rats.

Amiloride-sensitive Cation Channel, Neuronal proteins have two transmembrane domains (TM1 and TM2), a large extracellular loop and the N and C termini in the intracellular space (Chen et al. 1998). The proteins associate into homo- or heterotetramers to form cation transmembrane channels with differences in electrophysiological properties. One of these is lactic acid, also released during heart pain (Immke & McCleskey 2001a,b) and in intense muscle exercise periods, which is a situation encountered in human runners or swimmers when sprinting (Naves & McCleskey 2005). This should also apply to racing animals, such as horses and dogs, when cardiac and skeletal muscles turn to anaerobic metabolism at when the sustained speed is high. The characteristics of ACCN3 make it a good candidate for acidosis-related muscle pain sensitivity. Expression patterns and physical properties of the ACCN3 channel in rats identify ACCN3 as the detector of lactic acid (Naves & McCleskey 2005; Yagi et al. 2006), although it is also activated by other stimuli (Molliver et al. 2005). ACCN3 is specific to sensory neurons and is expressed in dorsal root ganglions (DRG) and absent in the brain (Waldmann et al. 1997a). Knockout mice for ACCN3 experience alterations in sensory perception, and this gene is postulated to modulate moderate- to high-intensity pain sensation (Price et al. 2001; Chen et al. 2002).

Similarly, ACCN1 mutations in the MDEG1 gene, which is an ACCN1 var 2 mammalian isoform, modify pH dependence, inactivation kinetics and amiloride sensitivity of this channel (Champigny et al. 1998). ACCN1 isoform 1 modulates the kinetics and pH sensitivity of ACCN2 and ACCN3 (Lingueglia et al. 1997; Yagi et al. 2006), and genetic alterations may also contribute to sensory variability. Mutations changing amino acids in the first pre-transmembrane region of ACCN1 and ACCN3 have been shown to affect the Na+/Ca+ selectivity and pH dependence of the ion channels (Coscoy et al. 1999).

Much work remains to be performed to elucidate the actual mechanisms by which the acid-gated channels function. However, when all the evidence is taken together, it appears that ACCN3 might be the preferential candidate for ischaemic pain, with a possible modulation of its functional properties by ACCN1.

In horses, the main cause of metabolic acidosis observed during both high-intensity and short-distance racing comes from anaerobic oxidation of glucose to lactic acid. In addition, plasma volume decreases because water moves from the vasculature to the intracellular and interstitial spaces at the onset of intense exercise (Hyyppa & Poso 1998). Thus, one may assume that lactic acidosis hampers horse performance in short-distance flat racing and that ACCN functional heterogeneity can influence this individual variability. Here, we have investigated the genomic structure of the ACCN1 and ACCN3 genes in the horse and have explored expression in different tissues. Transcript structures were also analysed for single nucleotide polymorphism (SNP) identification in different breeds and for alternative splicing.

Materials and methods

Gene structure analysis

Total RNAs were extracted from different tissues using the RNA Now procedure (Biogentex). Reverse transcriptions (RT) were performed on 5 μg of total RNA using the Superscript First Strand Synthesis System (Invitrogen) following the manufacturer’s instructions. RT on horse RNAs were performed using oligonucleotides (dT)-18 followed by PCR using the GoTaq® Flexi DNA Polymerase and reaction buffer (Promega).

The 3′ ends of the horse ACCN1 and ACCN3 cDNA were isolated using 3′RACE kit (Invitrogen). The 5′ ends of the horse ACCN3 cDNA were also cloned using 5′RACE kit (Invitrogen). We cloned and sequenced the most expected 5′ end by comparison with other species using proximal contiguous forward primers.

The cDNA sequencing of ACCN1 and ACCN3 was performed after RT-PCR amplification using respectively four and three primer pairs designed to amplify overlapping fragments 500–600 bp long covering all the exons of the genes (Table 1). All oligonucleotides of this study were designed using the primer 3 software ( The PCR amplification reactions were optimized and carried out in a PTC-100 (MJ-Research) using the following cycling conditions: 94 °C for 5 min followed by 35 cycles of (94 °C for 30 s, annealing temperatures (Table 1) for 30 s, 72 °C for 2 min) and 72 °C for 5 min. The resulting PCR products were separated on a 2% agarose gel, purified using Wizard SV Gel and PCR Clean-Up System (Promega) and sequenced by Qiagen sequencing services. The resulting sequences were compared to the horse genome sequence in the NCBI database by means of the blastn software ( to deduce the intron/exon structure of the ACCN1 and ACCN3 genes. All ACCN1 and ACCN3 genomic and cDNA sequences generated during the course of this study have been submitted to GenBank databases (HM467835, HM467836, HM467837, HM567408 and HM567409).

Table 1.   Primers used for RT-PCR (ACCN1_cDNA and ACCN3_cDNA) and semi qRT-PCR (in bold), for genomic amplification (ACCN1_Ex and ACCN3_Ex).
Primer namePrimer sequenceAnnealing temperature (°C)Exons


All exons of the ACCN1 and ACCN3 genes were amplified using primers in the introns on both sides of each exon. The PCRs were performed on genomic DNA extracted from the blood of a panel of 15 horses among four breeds (French trotter, French saddlebred, Thoroughbred and Norman cob) using the Genisol™ Maxiprep kit (ABgene) according to the manufacturer’s instructions. PCRs were optimized and DNA products were analysed as for the gene structure analysis. The primers used are listed in Table 1. The resulting sequences were introduced into the novosnp software (Weckx et al. 2005) to find SNPs in the amplified sequences.

Alternative splicing and expression of the ACCN1 and ACCN3 gene

RNAs were isolated from different tissues (skeletal muscle, heart, liver, spleen, lung, kidney and spinal cord) in one adult healthy horse. Semi-quantitative RT-PCR was conducted on these RNA tissue samples from five adult horses for ACCN1 and ACCN3 mRNAs. RT-PCR amplifications for splice variant determination were performed on several tissue RNA samples (skeletal muscle, heart, liver, and spinal cord). The primers used are presented in Table 1. PCRs were optimized and DNA products were analysed as for the gene structure analysis. The results of semi-quantitative RT-PCR were normalized with the GAPDH horse gene. The PCR amplifications used the following cycling conditions: 94 °C for 5 min followed by 28 cycles of (94 °C for 30 s, annealing for 30 s (temperatures in Table 1), 72 °C for 2 min) and 72 °C for 5 min.

A comparison of the tissue expression of the two ACCN mRNAs was made with the non-parametric Kruskal–Wallis one-way analysis of variance (Kruskal & Wallis 1952). It is an extension of the Mann–Whitney U-test for three or more groups. The test does not assume a normal distribution population but does assume an identically shaped and scaled distribution for each group, except for any difference in medians.

Promoter sequence analysis

The matinspector (Search transcription factor binding sites) program, part of the genomatixsuite (, was used for in silico proximal promoter analysis. About 1 kb upstream of the transcription start site (TSS) of ACCN1 was selected in the human, equine, canine, murine, and bovine species. For ACCN3, the sequences came from the Genomatix database. Stringent analysis parameters (matrix similarity = 0.05) giving a score >0.80 were used to select the most likely conserved transcription factors. The EMBOSS CpGPlot ( was used to detect CpG islands on the same sequences.


Sequence of the ACCN1 and ACCN3 cDNAs

The length of the ACCN1 ORF is 1539 nucleotides, with a predicted protein of 512 amino acids. Sequence analysis of this equine cDNA shows an identity of more than 92% compared to those of other mammalian species (Table 2 and Appendix S1). The ACCN3 ORF length is 1602 nucleotides, producing an expected 534 amino acid protein. The sequence identity between the equine ACCN3 cDNAs and other mammalian species is more than 83% (Table 2 and Appendix S2).

Table 2.   Length of ACCN1 and ACCN3 5′ and 3′UTR and ORFs, and percent identities in different species.
ACCN1Equus caballusHomo sapiensMus musculusBos taurus (predicted)Canis familiaris (predicted)
ORF (nt)1539153995%153992%153992%153993%
ORF (aa)51251299%51299%51299%51299%
3′UTR84187577%89273%86075% 336 (inc)91%
Total length26542688 2643 2570 2153 
ACCN3Equus caballusHomo sapiens (var 2)Mus musculusBos taurus (predicted)Canis familiaris (predicted)
ORF (nt)1602159688%159383%159988%160889%
ORF (aa)53353286%53085%53287%53583%
Total length21102088 1942 1708 1866 

Sequencing clones from the 3′ ends of these two cDNAs does not reveal any multiple polyadenylation sites. Cloning and sequencing the 5′ extremity of ACCN1 and ACCN3 do not present any alternative initiation transcription sites. Despite the use of several 5′ primers designed in the horse, we were unable to amplify the exon homologous to the human ACCN1 var 1 by RT-PCR.

ACCN1 and ACCN3 gene and ORF lengths and sequence identities compared between species and results are reported in Table 2. The ACCN1 gene structure is highly conserved, with the same length of each coding exon between species. For the ACCN3 gene, the situation is more complex, with small differences in the ORF length. There are two species that have one less amino acid (human and bovine) and one species that has three amino acids less (mouse), while the dog has a protein two amino acids longer. These differences within the coding sequence are principally localized in exon 4 of the human and mouse sequences. In the bovine and canine, sequence differences take place in exons 4 and 9 for the bovine sequence and in exons 5 and 6 for the canine sequence in comparison with the horse genome.

Structure of the ACCN1 and ACCN3 genes

The precise positions of the intron–exon junctions were deduced from the sequencing of the equine ACCN1 (ECA11) and ACCN3 (ECA4) mRNA sequences and the whole genomic sequence (Ecab2). The ten exons of the ACCN1 gene were identified (with the exception of the human exon 1b) as were the 11 exons of the ACCN3 gene. Each junction was delimited by the GT – AG crucial sequence at the beginning and the end of each exon (HM567408 and HM567409).

Polymorphism of ACCN1 and ACCN3 genes

Single nucleotide polymorphisms were located in the coding region of the cDNA of both genes by sequencing. All exons were sequenced on a panel of horses from different breeds. Ten SNPs were detected for ACCN1, five in the coding region and five in the non-coding region (Table 3). The SNPs in the coding region do not introduce any amino acid changes. In contrast, all three SNPs identified in the coding region of the ACCN3 gene introduce changes: an Alanine to Threonine at position 439, a Lysine to Isoleucine at position 569 and finally an Asparagine into Lysine at position 582 of the protein sequence.

Table 3.   Position of single nucleotide polymorphisms (SNPs) on the cDNA sequences of ACCN1 (HM467836) and ACCN3 (HM467837).
5′ UTRc.-125G>TExon 1 
ORFc.237G>Ac.439G>AExon 1p.Ala147Thr
c.756C>TExon 3c.582C>Ap.Asn194Lys
c.861G>TExon 4 
3′ UTRc.*434G>AExon 10

Alternative splicing of ACCN1 and ACCN3 genes

The expression of the human ACCN1 gene is reported to undergo a unique splicing involving only exons 1a and 1b. In fact, these two exons could not be found together in the transcripts of the same individual. In the horse muscle, RT-PCR could identify the homologous human exon 1a, but we were unable to identify exon 1b.

The situation is more diversified for ACCN3, where cDNA sequencing detects three alternative transcripts in muscle. One transcript contains all 11 exons, the second lacks the entire 72- bp exon 9 and the third contains an unspliced intron 8 (97 bp) in addition to the usual 11 exons. The lack of exon 9 in the second transcript does not change the frame, but it would be expected to shorten the protein by the small 24 amino acid domain encoded by exon 9. The last alternative transcript presents a shift of the open reading frame, with a new stop codon 96 bp further on, which would result in a longer protein of 597 amino acids with a different C-terminal. These expectations would only stand if these two cDNAs were translated, but we have no evidence that they do not come from an immature mRNA. We thus only report the exon 9 deletion in Fig. 1. No tissue-specific transcript was recognized in our RT-PCR amplifications of the tissues analysed.

Figure 1.

 Alternative splicing in the ACCN3 gene.

ACCN1 and ACCN3 gene expression

Expression analysis of ACCN1 and three genes was performed on different tissues (skeletal muscle, heart, liver, spleen, lung, kidney and spinal cord) by semi-quantitative RT-PCR on five adult horse RNAs. The results showed significant differential expression of the ACCN1 gene in the spinal cord tissue with the Kruskal–Wallis test (H = 18.578; P = 0.009) (Fig. 2).

Figure 2.

 Expression of ACCN1 and ACCN3 cDNAs in different horse tissues relative to a housekeeping gene (GAPDH).

Promoter sequence analysis

No TATA or CAAT box could be identified in the promoters of the ACCN1 and ACCN3 genes. EMBOSS CpGPlot detected a 200–300 bp-long CpG island in ACCN1 of all five species. No CpG island could be found for ACCN3. Several potential transcription factor (TF) sites were detected in the promoter regions of the five species, but only those present in three of five species are reported (Fig. 3). For ACCN1, five TF sites were observed within 100 bp downstream of the TSS (SNAP, HNF6, HOXF, CHRF and FKHD) and one (CDXF) <50 bp upstream of the TSS. It is also possible that one of the two HNF1 sites is present in the five species but a gap in the canine sequence makes it only suggestive. The second HNF1 site is only present in horses and humans, while an E2FF site is present in three species. The TF sites are less conserved in the ACCN3 promoter because none are present simultaneously in the five species. Our criteria detected five types of transcription factor sites (GLIF, IRFF, HAML, HAND and NR2F), which were all different from those present in the ACCN1 promoter.

Figure 3.

 Schematic representation of potential transcription factors in the promoter sequence of ACCN1 and ACCN3 genes.


The present study describes the structure of two horse genes that may be involved in pain production in horses submitted to high-intensity training. The horse ACCN1 and ACCN3 genes have the same number of exons as their vertebrate homologues, but we were unable to identify the horse counterpart of the human ACCN1 exon 1b. BLAST analysis of this human exon sequence finds a hit on the horse genome sequence in the 5′ region of the ACCN1 gene. Nevertheless, this human 1581 bp sequence shows 85% identity with the first 548 bp and more than 95% with the 357 last bp in the horse DNA, leaving a gap in the middle. This gap could not be detected in the equine traces database by directly blasting the corresponding human sequence or through blasting its two flanking sequences. It thus seems that this sequence is presently unresolved in the equine genome, appearing as an undetermined sequence and a 545 bp non-homologous sequence.

The ACCN1 gene showed a very high conservation between mammalian species at both the nucleotide and amino acid levels, with the same number of base pairs in the exons and the same protein length. In contrast, the ACCN3 equine protein is 1 (human) and 3 (murine) amino acids longer and 2 amino acids shorter than in canines. A careful analysis of the sequences of the ACCN1 and ACCN3 genes in different species shows that the length of introns and exons of the ACCN1 gene is very similar. The ACCN3 gene is more variable, especially in exon 4 (Table 2).

The three SNPs identified in the coding region of ACCN3 modify the amino acid sequence (p.Ala439Thr, p.Asn569Ile and p.Asn582Lys). All these amino acid substitutions change the polarity of the variants, as well as the charge being modified by the third SNP. It is not known whether these modifications affect the protein activity, but it should be noted that these mutations are not located in the two transmembrane domains nor in the glycosylation or phosphorylation sites when compared to the human protein (Human Protein Reference Database:

Concerning alternative transcripts, the deletion of ACCN3 exon 9 removes a region between amino acids 465 and 488 that contains a potential phosphorylation site in position 478. Retention of intron 8 introduces a frameshift, modifying the C-terminal sequence from position 464 and deleting two potential phosphorylation sites in positions 478 and 493. A new stop codon appears at position 2199 followed by a polyadenylation site 1179 nt downstream. However, it is possible that this transcript represents an immature mRNA of this gene.

The existence of several alternative ACCN3 transcripts (exon skipping of exon 9 and retention of intron 8) is not surprising. Four types of alternative splicing (exon skipping, partial 3′ or 5′ exon alternative splicing and intron retention) have been described (Kim et al. 2007, 2008). Rearrangements such as alternative splicing may affect 40% of the genes in humans and chickens and 30% in rats and mice. Their distribution is around 35–45% for exon skipping, 20–25% for each 3′ or 5′ exon partial splicing and <4–10% for intron retention.

Amiloride-sensitive Cation Channel, Neuronal gene expression is well documented in the central and peripheral nervous system but knowledge is rather scarce for other tissues. In one such, it was shown that ACCN3 is expressed in 14 different tissues, while ACCN1 var 2 is only expressed in the nervous system (Babinski et al. 2000). In the horse, the expression studies have yielded similar results, although ACCN1 mRNA is more highly expressed in the spinal cord; it was also detected at low levels in other tissues. There is no possible interference with the expression of ACCN1 var 1, because this variant could not be detected in the horse.

The promoter sequence of the ACCN1 gene is well conserved in the 2–300 bp upstream of the TSS in the five mammalian species studied, highlighting the importance of this sequence in the regulation of the expression of this gene (Xia et al. 2003). However, it should be noted that the transcription factors identified in silico by matinspector in this and other studies are different, which may be because of the ever growing knowledge of sequence–function relationship data, and thus the evolution of the transcription factor concept. The promoter sequences of the ACCN3 gene in the five species are less conserved, which explains the low level of conservation of transcription factors and precludes a simple common regulation of both genes.

This study brings experimental data to the predicted genomic structure of the ACCN1 and ACCN3 genes in the horse, as well as to their tissue expression, and opens new areas for further investigation of the role of ACCN polymorphisms in electrophysiological properties of cation channels and possibly also in horse pain.


This work was partly supported by a grant from the Haras Nationaux. Breeders and Haras Nationaux staff are greatly acknowledged for their help and constant interest in this work. We thank Dr Laurent Schibler for his advice during the promoter sequence analysis with matinspector and W. Brand-Williams for reviewing the manuscript.

Conflicts of interest

The authors have declared no potential conflicts.